We've sent a verification link by email
Didn't receive the email? Check your Spam folder, it may have been caught by a filter. If you still don't see it, you can resend the verification email.
Started October 28th, 2006 · 7 replies · Latest reply by Bram 17 years, 9 months ago
I've noticed that if I "overuse" noise reduction, the sample starts sounding as though it's being fed through a flanger*. Is there an explanation for this that can be understood by someone without a pro-level background in audio engineering or the like?
*or "burbling", "squelchy", "tinny", etc. -- I've been googling around for this and have found different ways of describing the sound, but haven't been able to find out why it happens.
Many thanks...
I know a lot about digital signal processing, you would have to specifically have done some noise reduction algorithms to know a good explanation of why the algorithms have that kind of filtering characteristic that you mention.
.milan
I've noticed that if I "overuse" noise reduction, the sample starts sounding as though it's being fed through a flanger*. Is there an explanation for this that can be understood by someone without a pro-level background in audio engineering or the like?*or "burbling", "squelchy", "tinny", etc. -- I've been googling around for this and have found different ways of describing the sound, but haven't been able to find out why it happens.
Many thanks...
This effect you described sounds very similar to what we can get with some softwares that creates sounds from pictures (like using a sonagram on the "wrong" way), we just hear soft frequencies bands everywhere, I think we can't really call that "flanger", it's more something about filtering, but anyway it's often nice to hear.
Nurykabe
I think we can't really call that "flanger", it's more something about filtering, but anyway it's often nice to hear.
As for an attempt at explaining briefly why it happens... the noise reduction algorithm is trying to establish how much of each frequency to filter out, based on the expected noise... but because noise by definition covers a lot of frequencies, when you overdo the reduction, you start taking out more and more noticeable portions of the original sound as well.
I hope that's some use. I don't know as much about signal processing as I'd like, but that's the rough gist I guess...
noise reduction tries to selectively remove certain frequencies from audio, but trying to do that can sometimes create artifacts. This is very common in FFT or spectral processing. WHere exactly the artifacts come from depends on the noise reduction algorithm. But you could safely say that the algorithm is deleting too much information, thus messing up... everything.
Nurykabe is quite right. The sounds do sound like sound->image->sound software. This is because the software that the sound=>image=>sound things also use spectral processing and you get the same type of errors.
If you want to experiment with these kind of "errors", I advise you to check out MDSP's FFT tools;
http://mdsp.smartelectronix.com/ffttools/
and Plogue Bidule's spectral processing;
http://plogue.com
- bram
Oh ! These tools are good, thanks for the links, Bram. By the way, is there any FFT software that really works well ? I mean, that lets you, for example, graphically edit a spectrogram. I heard of "audiosculp" on mac, but I don't know if there is one of the same kind on PC.